Editor's Page Cite This: Organometallics 2019, 38, 603−605
pubs.acs.org/Organometallics
How To Make Your Computational Paper Interesting and Have It Published
■
Organometallics 2019.38:603-605. Downloaded from pubs.acs.org by 91.243.190.120 on 02/11/19. For personal use only.
I
THE SCIENTIFIC PROBLEM A paper whose scientific hypothesis is restricted to “We computed the mechanism that the experimentalist proposed, and it seems to work okay.” is not very exciting to read. Use computations to study an interesting problem, and state it clearly in the text. For example, “Why does catalyst1convert ketones, but not aldehydes, which should be more reactive?”7 Choose problems where theory can provide answers that experimentalists would love to have but cannot access easily.
n 1987, the late German chemist Peter Hofmann of the University of Heidelberg wrote: “We still have a long way to go until a computation will be able to compete with or to substitute a lab experiment... one could then conclude that theoretical work in this fieldnot being quantitatively reliable anyhowis rather useless, except for the purpose of keeping theorists busy.” However, Hofmann continued: “If applied properly and with their limitations in mind, methods of various levels of sophistication can all contribute their part to a basic understanding of organometallic systems.”1 Without doubt, the past 30 years have shown the usefulness of computational methods to provide information about the properties of molecules, including the activities and selectivities of organometallic systems.2−4 Since the late 1990s, the preferred method of computational chemists has been density functional theory (DFT).4,5 A search of papers containing DFT in Organometallics shows an increase from 113 papers in 2000 to 379 in 2012 (Figure 1).6
■
THE PRESENTATION An abstract or discussion should be easy to follow. “Our calculations show that Int1 becomes Int2 via TSab_II, whereas the pathways through Int7 and TS19 are clearly disfavored.” is unclear. Instead, try “The preferred pathway proceeds via a carbene intermediate.” Give the exact experimental conditions in your discussion: solvent, temperature, and mol % catalyst. The reader wants to know what you are trying to model. Refer to those who proposed the mechanism, including in figure captions. Talking about Figures. Show clear pictures of relevant transition states (TSs), but instead of ball-and-stick models, consider ChemDraw figures, which are visibly more accessible to experimentalists (Figure 2). Avoid overloading energy
Figure 1. Approximate number of articles containing DFT in Organometallics.6
Organometallics is now receiving an ever-increasing number of manuscripts that report DFT calculations in whole or in part. Unfortunately, computational manuscripts sometimes are written in a way that is not very appealing to experimental readers (or reviewers and editors). Furthermore, computed results often are not sufficiently correlated with experimental data, as noted in one of our recent Tutorials (DOI: 10.1021/ acs.organomet.8b00456). The consequence may be a rejection of the manuscript. How can you increase the likelihood that your computational study is publishable in Organometallics? Three things are essential:
Figure 2. (a) Ball-and-stick versus (b) a ChemDraw figure of the same TS. Axial ligands omitted, adapted from the author’s own work (DOI: 10.1021/om1009507). Copyright American Chemical Society 2011.
profiles with 4 or 5 different pathways. A mechanistic figure should not require half an hour of contemplation (Figure 3), so let an experimental colleague evaluate your figures and text before submission. Give the interested reader the option to visualize structures by providing them in a separate Supporting Information (SI) file that can be opened with Mercury or other applications (see ref 8). Most Important, Keep It Short. You probably would not be excited to read 15 pages about experiments that failed. Do not expect that an experimentalist wants to read 15 pages about computations that are not relevant. For nonpreferred
• The scientific problem should be intriguing. • The presentation should be appealing. • The computed results should be validated. The following sections provide some (personal) advice on how to make a computational paper intriguing, appealing, and validated. The main focus is on mechanistic studies. © 2019 American Chemical Society
Published: February 11, 2019 603
DOI: 10.1021/acs.organomet.8b00942 Organometallics 2019, 38, 603−605
Organometallics
Editor's Page
aldehydes, then your mechanism needs to reproduce that result. When you compute enantiomeric excess (ee), estimate the error in kcal/mol. If you wanted 33% ee (R), but got 10% ee (S), then you may be worried, but the error is only 0.5 kcal/ mol (298 K). If you wanted 99.5% ee (S) and got 90% ee (S), then you may not be worried, but maybe you should be, as the error is 1.8 kcal/mol. This may indicate your TS or mechanism is incorrect. Some computational papers make predictions that have not yet been tested experimentally. This should be done in combination with studying something known. If your computed results about known ligand A match experiment well, your predictions about novel ligand B are more credible. The most compelling papers may be those where computations and experiments are employed together.2,12 This provides the opportunity to use theory to immediately rationalize interesting experimental results and to use experiment to immediately validate interesting computational predictions.
Figure 3. (a) Incomprehensible figure by the author that she is not very proud of today (DOI: 10.1021/om400755k). Copyright American Chemical Society 2013. (b) Comprehensible figure.9
mechanisms, refer briefly to the results, but place the details in the SI.
■
METHODS AND MODELS The model should be as close as possible to the real system. If in experiments, 4-chlorostyrene was converted, but styrene did not react, then do not base your mechanistic investigation on styrene. In addition, choose an appropriate DFT method. Today, this normally means including dispersion corrections.10,11 Do not use DFT for mechanisms it cannot handle. Be careful if many ligands enter or leave; this is difficult for conventional DFT to describe accurately. Therefore, use a sufficiently large basis set and correct for basis set superposition errors.10,11 Also, avoid using static DFT for a reaction with many loosely interacting components. For example, if in your model, the catalyst, substrate, additive, and solvent interact through weak forces, such as dispersion interactions or hydrogen bonds, then many different supramolecular arrangements can be formed. Picking just one of these as an energy reference can introduce large errors; instead, conformational sampling and averaging over all arrangements would be required (DOI: 10. 1021/acs.organomet.8b00456). For the same reason, do not include explicit solvent, unless it is strongly bound to a complex.
■
PROPOSED REVIEWERS A computational study is often based on a specific experimental paper. Surprisingly, computational authors sometimes indicate the experimentalist as a nonpreferred reviewer. Editors can disregard this request. If you instead indicate the experimentalist as a preferred reviewer, then it shows that you are not worried to have the results scrutinized by an expert on this system. If he/she finds the work sound and convincing, this is a quality stamp and increases the likelihood that your results will impact future experimental setups. Is not that a main goal of any computational study? With this hopefully helpful (and friendly) advice, Organometallics is looking forward to receiving your intriguing, appealing, and validated computational study.
Kathrin H. Hopmann
■
■
Hylleraas Centre for Quantum Molecular Sciences, Department of Chemistry, University of TromsøThe Arctic University of Norway, N-9037 Tromsø, Norway
AUTHOR INFORMATION
ORCID
RESULTS Be critical of your results! Is the computed barrier meaningful? At 298 K, a barrier should be below 25 kcal/mol (DOI: 10. 1021/acs.organomet.8b00456). Also, if it is below this value, then make sure to test mechanistic alternatives. Always discuss in the manuscript if the results really are in agreement with experiment. If an additive speeds up the reaction by a factor of 4, then this corresponds to a lowering of the barrier by 0.8 kcal/mol (298 K). If your computations predict that the additive lowers the barrier by 9 kcal/mol, then the trend may look right, but this is a factor of 4 million and hence not in agreement with experiment. If you locate an intermediate that has a low energy, then do not ignore it. If it is easy to form, then it needs to be considered as an on- or off-cycle species. If this makes your barriers too high, then your mechanism may be incorrect.
Kathrin H. Hopmann: 0000-0003-2798-716X Notes
Views expressed in this editorial are those of the author and not necessarily the views of the ACS. Biography
■
VALIDATION This is most important: always validate the final mechanism. How? (i) Compare computed intermediates to experimentally observed intermediates. (ii) Show that the proposed mechanism can reproduce substrate preferences and selectivities. If in experiments ketones are better substrates than
Kathrin H. Hopmann is Associate Professor in Computational Chemistry at UiTThe Arctic University of Norway and an 604
DOI: 10.1021/acs.organomet.8b00942 Organometallics 2019, 38, 603−605
Organometallics
Editor's Page
Associate Editor for Organometallics. Her research focuses on using computational methods to elucidate the detailed nature of chemical reactions, in particular mechanistic aspects that control the selectivity in organometallic reactions.
■ ■
ACKNOWLEDGMENTS I would like to thank all authors, reviewers, and editors for their tireless contributions to Organometallics. REFERENCES
(1) Hofmann, P. Organometallic Reactivity and Applied Quantum Chemistry - Some Aspects of CH-Activation. In Organometallics in Organic Synthesis; de Meijere, A., tom Dieck, H, Eds.; Springer: Berlin, Heidelberg, 1987. (2) Lin, Z. Interplay between Theory and Experiment: Computational Organometallic and Transition Metal Chemistry. Acc. Chem. Res. 2010, 43, 602−611. (3) Sperger, T.; Sanhueza, I. A.; Kalvet, I.; Schoenebeck, F. Computational Studies of Synthetically Relevant Homogeneous Organometallic Catalysis Involving Ni, Pd, Ir, and Rh: An Overview of Commonly Employed DFT Methods and Mechanistic Insights. Chem. Rev. 2015, 115, 9532−9586. (4) Hopmann, K. H. Quantum chemical studies of asymmetric reactions: Historical aspects and recent examples. Int. J. Quantum Chem. 2015, 115, 1232−1249. (5) Jones, R. O. Density functional theory: Its origins, rise to prominence, and future. Rev. Mod. Phys. 2015, 87, 897−923. (6) Search performed at https://pubs.acs.org/search/advanced with search string “Density Functional Theory” OR DFT. Reviews, editorials, and corrections were excluded from the results. Analysis of the search results for 2000 and 2003 reveals that about 10% of returned articles referred to DFT results by others, but the authors did not report own computations. Numbers shown in Figure 1 are therefore approximate. (7) This and all text examples given here are invented by the author and do not originate from real manuscripts. Any similarity is purely accidental. (8) This is explained in the Organmetallics Author Guidelines: https://pubs.acs.org/paragonplus/submission/orgnd7/orgnd7_ authguide.pdf. (9) Based on similar figures from (a) Obst, M.; Pavlovic, Lj.; Hopmann, K. H. Carbon-Carbon bonds with CO2: Insights from Computational Studies. J. Organomet. Chem. 2018, 864, 115−127. (b) Ostapowicz, T. G.; Hölscher, M.; Leitner, W. Chem. - Eur. J. 2011, 17, 10329−10338. (10) (a) Kruse, H.; Goerigk, L.; Grimme, S. Why the Standard B3LYP/6-31G* Model Chemistry Should Not Be Used in DFT Calculations of Molecular Thermochemistry: Understanding and Correcting the Problem. J. Org. Chem. 2012, 77, 10824−10834. (b) Hansen, A.; Bannwarth, C.; Grimme, S.; Petrović, P.; Werlé, C.; Djukic, J.-P. The Thermochemistry of London Dispersion-Driven Transition Metal Reactions: Getting the ‘Right Answer for the Right Reason’. ChemistryOpen 2014, 3, 177−189. (11) Hopmann, K. H. How Accurate is DFT for Iridium-Mediated Chemistry? Organometallics 2016, 35, 3795−3807. (12) Camasso, N. M.; Canty, A. J.; Ariafard, A.; Sanford, M. S. Experimental and Computational Studies of High-Valent Nickel and Palladium Complexes. Organometallics 2017, 36, 4382−4393.
605
DOI: 10.1021/acs.organomet.8b00942 Organometallics 2019, 38, 603−605
